The Psychology of Risk
Paul Slovic
August 10,2009
Some Questions Briefly Addressed:
1.How do people think about risk?
2.What factors determine the perception
of risk and the acceptance of risk?
3.How do we make decisions in situations
of risk?
Overview
1.Judgment under Uncertainty:Intuitive Statistics
2.Studies of Perceived Risk
-experts’ and laypersons’ perceptions
-the social amplification of risk
-intuitive toxicology
-cultural cognition,worldviews,and risk perception
-numeracy and risk perception
3.Risk and Trust
Overview,continued
4.Risk as Analysis vs. Risk as Feeling
-the importance of affect
-experiential vs.analytic thinking
5.The Affect Heuristic
-probability neglect
-insensitivity to large numbers
6.Hunches and Risk Perception
Risk Perception
Lessons from Risk Perception Research
2.Every hazard is uniquely understood and
evaluated in terms of its characteristic qualities.
Risk is Multidimensional
Qualitative Risk Concerns
•
•
•
•
•
•
•
•
•
•
Voluntary – Involuntary
Chronic – Catastrophic
Common – Dread
Certainly not fatal – Certainly fatal
Known to exposed – Not known to exposed
Immediate – Delayed
Known to science – Not known to science
Not Controllable – Controllable
New – Old
Equitable – Not equitable
Voluntary
Involuntary
Chronic
Catastrophic
Common
Certain not fatal
Nuclear power
X-rays
Known to exposed
Dread
Certainly fatal
Not known to exposed
Delayed
Immediate
Known to science
Not known to science
Not controllable
Controllable
New
Old
1 2 3 4 5 6 7
Mean rating
Acceptance of Risk Tends to be Reduced if:
•The hazard is new or unfamiliar
•exposure to the hazard is involuntary
• the risk is not under one’s control
• the risk evokes feelings of dread
• the outcomes are catastrophic
• the benefits of an activity are not highly visible
or not fairly or equitably distributed among those
who bear the risks.
Acceptance of Risk Tends to be Reduced if:
• the risk is posed by human failure as opposed
to natural causes
• the potential harms are genetic and/or delayed
in time
• the risk is perceived as not well known to
science or to those who might be harmed
Lessons from Risk Perception Research
3.Perceptions of Risk have Impacts
-the social amplification of risk
The Social Amplification of Risk
Individual risk perceptions and cognitions,
interaction with social and institutional forces,
can trigger massive social and economic
impacts due to
• response to events (even “small” incidents)
• stigma effects
The Social Amplification of Risk
Media coverage
E
Ec1
Ec2
Ec3
.
.
.
Ecn
Other technologies
Industry
Company
Interpretation of
E/ signals
Victims
Loss of trust
Imagery & affect
Event
Special information
characteristics and Interpretation
Litigation
Stigma
loss of sales
community
opposition
investor flight
----------------------
Event
Regulatory
constraints
Spread of impact
Type of impact
(company level)
A preliminary model of social amplification of risk and stigma impacts.
Development of the model will require knowledge of how the characteristics (Ec)
associated with a hazard event interact to determine the media coverage and
the interpretation or message drawn from that event. The nature of the media
coverage and the interpretation is presumed to determine the type and
magnitude of ripple effects.
Source: Kasperson et al. (1988).
Accidents Are Signals
1.
The perceived seriousness of a mishap, the media coverage it
gets, and the long-range costs to the responsible company,
industry, or agency are determined by the mishap’s signal
value
2.
Signal value reflects perception that the event provides new
information about the likelihood of similar or more destructive
future mishaps
3.
High signal events: Bhopal, Chernobyl, Challenger
“What truly grips us in these accounts [of disaster] is not so
much the numbers as the spectre of suddenly vanishing
competence, of men utterly routed by technology, of failsafe systems failing . . . And the spectre haunts us
because it seems to carry allegorical import, like the
whispery omen of a hovering future.”
The New Yorker, February 18, 1985
4.Lessons from risk perception research
People’s ideologies and worldviews strongly
influence their perception and acceptance
of risk.
Worldviews
Culture determines salient values
•
Douglas & Wildavsky (1982):
Every cultural group attends to some risk and ignores other to maintain a
particular way of life.
- Hierarchist (support superior/subordinate social relations and detest civil
disobedience
- Individualists (support self-regulation, individual achievement and reward
and dislike social rules that constrain individual initiative)
- Egalitarians (support broad distribution of power and wealth and dislike
ranked role differentiation)
- Fatalists (see nature as capricious and thus uncontrollable)
If your community was faced with a potential shortage
for electricity, do you [strongly agree… strongly
disagree] that a new nuclear power plant should be built
to supply that electricity?
100%
80%
64%
Actual
Response
% agree
60%
47%
43%
35%
40%
20%
0%
Quartile 1
Low
Quartile 2
Quartile 3
Egalitarian worldview
Quartile 4
High
Intuitive Toxicology
(Kraus, Malmfors, & Slovic)
Intuitive Toxicology Subtopics
a) conceptions of toxicity, including the toxicity of natural vs.
synthetic substances,
b) effects of chemical concentrations,dose,and exposure on
perceptions of risk,
c) the value of animal studies for predicting toxicity in humans,
d) interpretation of evidence regarding cause-effect relationships
between exposure to chemicals and human health,
e) confidence in the ability of scientific methods to discover and
quantify the toxic effects of chemicals
f)
interpretations of scientific expressions commonly used in
reporting chemical risks,
g) attitudes toward chemical use & regulation.
Intuitive Toxicology — Main Result
Many people lack dose-response sensitivity for exposure to
chemicals that can produce effects that are dreaded, such as cancer
(high affect).
If large exposures are bad, small exposures are also bad.
High
Public
Toxicologists
Cancer
risk
Low
Low
High
Exposure
Small
probability
of harm
High
probability
of harm
Dose-response insensitivity is part of a
more general neglect of probability when
the outcomes are highly emotional(eg,
cancer,mad cow disease,terrorism).
Kai Erickson’s contamination model of perceived risk
and stigma may explain dose insensitivity as well as
the strong fear of nuclear and chemical accidents.
Erickson describes the exceptionally dread quality
of technological accidents that expose people to
radiation and chemicals in ways that
‘contaminate rather than merely damage; . . .
pollute, befoul, and taint rather than just create
wreckage; . . . penetrate human tissue indirectly
rather than wound the surface by assaults of a
more straightforward kind’ (p. 120).
Unlike natural disasters, these radiation and chemical
accidents are unbounded. Unlike conventional disaster
plots, they have no end.
‘Invisible contaminants remain a part of the
surroundings — absorbed into the grain of the
landscape, the tissues of the body, and worst of
all, into the genetic material of the survivors. An
‘all clear’ is never sounded. The book of accounts
is never closed’ (p. 121).
Erickson’s ‘contamination model’ may explain the
reaction of the public to exposures to carcinogens.
• Numerous studies have found that a high percentage
(60-75%) of people believe that if a person is exposed
to a chemical that can cause cancer, that person will
probably get cancer some day.
• A similarly high percentage believe that ‘exposure to
radiation will probably lead to cancer some day.’
• The belief that any exposure to a carcinogen is likely
to lead to cancer tends to coincide with the belief that
it can never be too expensive to reduce such risks.
Lessons from Risk Perception Research
5. Trust is critical and also fragile.
Risk
and
Trust
Importance of Trust
“Acceptance of any risk is more
dependent on public confidence in
risk management than on
quantitative estimates of risk”
C. Starr
TRUST is the Key to
Communication Problems
• If you have trust the path is smooth
• If you do not have trust, no form of
phrasing or presentation is likely to
be successful
Trust Is Fragile
“If you once forfeit the confidence of
your fellow citizens, you can
never regain their respect and
esteem”
A. Lincoln
Trust: The Asymmetry
Principle
It is far easier to destroy trust than to create it!
• Negative events are more sharply defined (accidents,
lies)
• Positive events are often fuzzy or indistinct
e.g., how many positive events are represented by
the safe operation of a nuclear power plant for one
day?
• Negative (trust-destroying) events outweigh positive
events
Results
Local board authority to close plant
Evacuation plan exists
On-site government inspector
Rewarded for finding problems
Responsive to any sign of problems
Effective emergency action taken
Local advisory board established
Public encouraged to tour plant
Mandatory drug testing
No problems for five years
Hold regular public hearings
Employees carefully trained
Conduct emergency training
Community has access to records
Serious accident is controlled
Nearby health is good
Monitor radioactive emissions
Employees informed of problems
Neighbors notified of problems
No evidence of withholding information
Contribute to local charities
Employees closely supervised
Try to meet with public
Managers live nearby
Operates according to regulations
No problems in past year
Record keeping is good
Don’t contribute to local charities
No public hearings
Little communication with community
Emergency response plans not rehearsed
Officials live far away
Poor record keeping
Accident occurs in another state
Accused of releasing radiation
Denied access to records
Employees not informed of problems
Delayed inspections
Public tours not permitted
Health nearby worse than average
Official lied to government
Serious accident is controlled
No adequate emergency response plan
Plant covered up problem
Employees drunk on job
Records were falsified
TRUST
DECREASING
60%
40%
TRUST
INCREASING
20%
0%
20%
40%
60%
Percent very powerful impact
Differential Impact of Trust-Increasing and Trust-Decreasing Events
Impact
Very small
1
Results
Very
powerful
2 ...6
7
The county medical
examiner reports
that the health of
people living near
the plant is worse
than the average
for the region
3.0
8.0
26.0
24.0
The county medical
examiner reports
that the health of
people living near
the plant is better
than the average
for the region
21.5
14.0
16.1
2.2
Impact
Very small
1
Very
powerful
2 ...6
7
The plant is found
to lack an adequate
emergency
response plan
1.0
1.9
19.4
35.0
There is an
emergency
evacuation plan for
the people living
near the plant
19.5
10.3
8.1
10.3
Part II
Risk As Analysis and Risk As Feelings
Risk As Analysis vs. Risk as Feelings
Analytic/
Deliberative
Experiential/
Affective
Process and Content in Two Cognitive Systems
Source: Kahneman, 2003
Risk primarily resides in us as a “gut feeling”.
Neural Economics
P. Read Montague (Baylor College of Medicine)
• Survival is about economic evaluation
• The brain is an economic evaluation engine
• The core of neural economics is the recognition that rapid, ongoing economic
evaluation is a central… function carried out by the nervous systems of mobile
creatures.
• Without some kind of internal currency in the nervous system, a creature would be
unable to assess the relative value of events like drinking water, searching for
predators, or chasing prey. The nervous system must estimate the value of each of
these actions and convert it to a common scale. Recent work has shown that
fluctuations in the delivery of the neurotransmitter dopamine may represent one such
common currency.
•
At the psychological level
this “common currency”
may be affect:
A valenced quality (e.g., goodness
or badness) associated
with a stimulus
High

Activities,
hazards, etc.


Risk


















Low
Low
High
Benefit
In the world, risk and benefit are positively correlated.
In people’s minds, they are negatively correlated.
High
Activities,
hazards, etc.


















Risk


Low
Low
High
Benefit
Relationship between risk and benefit in people’s minds
Figure 1 Perceived risks and benefits of nanotechnology and 43 other technologies, based on 503
responses to a national telephone survey. Source:Currall et al. 2006
The Affect Heuristic
How do I feel
about a
nuclear waste
repository?
+/–
Perceived
benefit
Perceived
risk
A model of the affect heuristic explaining the risk/benefit
confounding observed by Alhakami and Slovic (1994).
Judgments of risk and benefit are assumed to be derived
by reference to an overall affective evaluation of the
stimulus item.
Study 1: Risk and Benefit
Judgments under Time Pressure
 Time pressure reduces opportunity
for analytic deliberation, and
increases reliance on affect
 Prediction: Under time pressure
people are more likely to use the
affect heuristic to make
judgments.
Result: Time pressure increases the inverse relationship between risk and
benefit.
Study 2: Manipulating Affect by Providing Risk and Benefit
Information
Technique: provide information to change
overall impression, e.g., create a more
favorable affective evaluation of nuclear
power with info that it has high benefit.
Perceived risk should then decrease.
Nuclear Power
Affect
+
Information: benefit is
high
Inference: risk
is low
A Nuclear Power
Positive
Information
says “Benefit
is high”
Risk
inferred to
be low
C Nuclear Power
Negative
Information
says “Benefit
is low”
Risk
inferred to
be high
B Nuclear Power
Positive
Information
says”Risk is
low”
Benefit
inferred to
be high
D Nuclear Power
Negative
Information
says “Risk is
high”
Benefit
inferred to
be low
Model showing how information about benefit (A) or information
about risk (B) could increase the overall affective evaluation of
nuclear power and lead to inferences about risk and benefit that
coincide affectively with the information given. Similarly, information
could decrease the overall affective evaluation of nuclear power as
in C and D.
Source: Finucane et al. (2000).
Question:
We have just seen evidence that risks and benefits
are confounded in people’s minds.What might this
mean for risk communication and decision making?
My focus next will be the role of affect
in the processing of numerical information.
Strong Affect Overcomes Probability
Prices paid to
avoid electric
shock and
$20 penalty
$20
18
16
14
12
10
8
6
4
2
0
Money
Shock
1%
99%
Probability
Payment to avoid a chance of electric shock is not much affected by probability
Source: Rottenstreich & Hsee:
Money, Kisses, and Electric Shock: On the Affective Psychology of Risk.
Psychological Science, 2001
Intuitive Toxicology — Main Result
Many people lack dose-response sensitivity for exposure to
chemicals that can produce effects that are dreaded, such as cancer
(high affect).
If large exposures are bad, small exposures are also bad.
High
Public
Toxicologists
Cancer
risk
Low
Low
High
Exposure
Small
probability
of harm
High
probability
of harm
Terrorism and Probability Neglect
Cass R. Sunstein
The Journal of Risk and Uncertainty, 26(2/3); 121-136, 2003
• People are prone to . . . probability neglect, especially
when their emotions are intensely engaged.
Probability neglect is highly likely in the aftermath of
terrorism. People fall victim to probability neglect if
and to the extent that the intensity of their reaction
does not greatly vary even with large differences in
the likelihood of harm. When probability neglect is at
work, people’s attention is focused on the bad
outcome itself, and they are inattentive to the fact that
it is unlikely to occur.
• Implications for possible risks from chemicals.
Another Example of the Affect Heuristic
Probability and Relative Frequency in
Risk Communication
• Are they the same or different in
communicating risk?
e.g., 1% chance
vs.
1 out of 100
Work with John Monahan,Ellen Peters,& Don MacGregor.
RISK COMMUNICATION:
A patient – Mr. James Jones – has been evaluated for discharge from an
acute civil mental health facility where he has been treated for the past
several weeks. A psychologist whose professional opinion you respect has
done a state-of-the-art assessment of Mr. Jones. Among the conclusions
reached in the psychologist’s assessment is the following:
EITHER:
Patients similar to Mr. Jones are estimated to have a 20% probability of
committing an act of violence to others during the first several months after
discharge.
OR:
Of every 100 patients similar to Mr. Jones, 20 are estimated to commit an
act of violence to others during the first several months after discharge.
Question:
• If you were working
as a supervisor at this
mental health facility
and received the
psychologist’s report,
would you recommend
that Mr. Jones be
discharged from the
hospital at the present
time?
60%
41%
40%
21%
20%
0%
20%
probability
20 of 100
patients
Do not discharge
Patient Evaluation
10%
• Very few people are violent
• 10% = 1/10
• Probably won’t hurt anyone, though
•
•
•
•
•
1 out of 10
He could be the 1 out of 10
Some guy going crazy and killing people
The patient attacking someone
An act of violence
There has to be at least 1 in 10. Mr. Jones could very well be
that 1
How Should We Value the
Saving of Human Lives?
A normative model:
Every human life is of equal value
How Should We Value the
Saving of Human Lives?
Another normative model: Large losses threaten
the viability of the group or society
But our actions in the face of mass murder
do not follow either of these normative models.
Our feelings overide our analytic judgments!
A descriptive model of diminished sensitivity as N
grows large. All lives are not valued equally.
(psychophysical numbing)
Another descriptive model:The collapse of compassion.
Our capacity to feel (good or bad) is limited. Valuation
depends on feelings (the affect heuristic). Lack of
feeling (value) leads to inaction as large losses of life
occur in episodes of mass murder or genocide.
Hunches and Risk Perception
The United States military has spent billions on hardware, like signal
jamming technology, to detect and destroy what the military calls
improvised explosive devices, or I.E.D.’s, the roadside bombs that have
proved to be the greatest threat in Iraq and now in Afghanistan, where
Sergeant Tierney is training soldiers to foil bomb attacks.
Still, high-tech gear, while helping to reduce casualties, remains a mere
supplement to the most sensitive detection system of all – the human brain.
Troops on the ground, using only their senses and experience, are
responsible for foiling many I.E.D. attacks, and, like Sergeant Tierney, they
often cite a gut feeling or a hunch as their first clue.
Everyone has hunches – about friends’ motives, about the stock market,
about when to fold a hand of poker and when to hold it. But United States
troops are now at the center of a large effort to understand how it is that in a
life-or-death situation, some people’s brains can sense danger and act on it
well before others’ do.
The study complements a growing body of work suggesting that the speed
with which the brain reads and interprets sensations like the feelings in
one’s own body and emotions in the body language of others is central to
avoiding imminent threats.
“Not long ago people though of emotions as old stuff, as just feelings –
feelings that had little to do with rational decision making, or that got in
the way of it, “ said Dr. Antonio Damasio, director of the Brain and
Creativity Institute at the University of Southern California. “Now that
position has reversed. We understand emotions as practical action
programs that work to solve a problem, often before we're conscious of it.
These processes are at work continually, in pilots, leaders of expeditions,
parents, all of us.”
Moral intuition is like perception
• Visual perception is sophisticated
and remarkably accurate
• But perception is subject to
powerful systematic biases
The squares marked A and
B are the same shade of
gray.
• Intuitive judgment, like perception, is
sophisticated, accurate, and sometimes
very biased
• One of the problems was that a lack of hard data
prevented the input of more common sense
analysis . . .
• The overriding emphasis on data – the kind of
analysis that repeatedly produced “no flight safety
risk” assessments involving Columbia’s external
tank foam strike – paints NASA decision makers
into a corner and fails to take advantage of
engineering common sense that may not be
backed up by specific data . . .
• ‘intuition and hunch’ do not carry any weight.
They do in everyday decision-making. But when
it comes to formal decisions, hard data-numbers
are required
• . . .the underlying rules are that as engineers you
have to have ‘the number’” around which to base
assessments, . . . “That basically means every
flight becomes ‘data’ and that ‘concerns’ about an
anomaly are not data – so a successful flight with
an anomaly simply becomes data that say it’s safe
to fly.
How can we improve our intuitive skills?
• create proper learning environments
• recognize and avoid deceptive learning
environments
• recognize the strengths and weaknesses
of intuitive and deliberative processes
• restructure the task or environment to correct
known bias
Conclusions
• The analytic and experiential (affective) systems of
thought are exquisitely sophisticated and embody
the essence of human rationality.
• Both systems, however, can lead us astray.
• Each system needs the other for guidance.
• Effective use of hunches depends on the ability to
know when they work well and when they don’t.